NOUMENOpen app
NOUMEN

OpenAI · Efficient

OpenAI o4-mini high

Added Dec 4, 2025

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

The maximum/high version of the o4-mini model. The next generation of OpenAI models.

  • Reasoning
  • Tools
  • Structured output
Input
Text
Output
Text
Reasoning effort
low · medium · high
Discovery tier
Efficient

Pricing

USD per 1M tokens
Input
$1.10
Output
$4.40
Cache read
$0.55

Reference prices as published by the model provider. API example

Specifications

Context window
200K
Max output
100K
Average output (7 days)
75 tokens
Longer than 7% of 468 models

Side by side

OpenAI o4-mini high compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

OpenAI o4-mini highOpenAIThis modelGPT 6 LunaOpenAIRemove
ProviderOpenAIOpenAI
Discovery tierEfficientEfficient
AddedDec 4, 2025Sep 22, 2026
Limits
Context window200K1.05MBest
Max output100K128KBest
Average output75 tokens120 tokens
Pricing, USD per 1M tokens
Input$1.10$0.10Best
Output$4.40$0.50Best
Cache read$0.55$0.01Best
Cache write—$0.13
Performance
Throughput—296.3 tok/s
Time to first token—3.8 s
Request success (24h)—99.7%
Benchmark indexes
Intelligence Index16.729.9Best
Math Index90.7—
Capabilities
ReasoningYesYes
ToolsYesYes
VisionNoYes
Structured outputYesYes
PDF inputNoYes
Reasoning effortlow, medium, highnone, low, medium, high, xhigh, max

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • OpenAI o4-mini high
  • GPT 6 Luna

Intelligence Index

  • OpenAI o4-mini high16.7
  • GPT 6 Luna29.9

OpenAI o4-mini high tested at high reasoning effort

OpenAI o4-mini high: Better than 44% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Math Index

    Composite of the math evaluations

    • OpenAI o4-mini high: 90.7
    • GPT 6 Luna: —

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 40.5%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 10.2%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 1,097 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 1,263 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 13.8%

  • AA-LCRv1.1

    Reasoning across long documents

    • OpenAI o4-mini high: 61.0%
    • GPT 6 Luna: 78.3%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 18.3%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • OpenAI o4-mini high: 16.5%
    • GPT 6 Luna: 28.3%Top

  • GPQA Diamond

    Graduate-level science questions

    • OpenAI o4-mini high: 78.4%
    • GPT 6 Luna: —

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • OpenAI o4-mini high: 83.2%
    • GPT 6 Luna: —

  • IFBench

    Following precise output instructions

    • OpenAI o4-mini high: 68.7%
    • GPT 6 Luna: —

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 2.5%

  • SciCode

    Scientific computing in Python

    • OpenAI o4-mini high: —
    • GPT 6 Luna: 50.9%

  • LiveCodeBench

    Recent competitive programming problems

    • OpenAI o4-mini high: 85.9%
    • GPT 6 Luna: —

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • OpenAI o4-mini high: 90.7%
    • GPT 6 Luna: —

  • AIME

    Competition mathematics

    • OpenAI o4-mini high: 94.0%
    • GPT 6 Luna: —

  • MATH-500

    Competition math problems

    • OpenAI o4-mini high: 98.9%
    • GPT 6 Luna: —

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • OpenAI o4-mini high: 15.2%
    • GPT 6 Luna: —

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • OpenAI o4-mini high: 55.6%
    • GPT 6 Luna: —

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • OpenAI o4-mini high: 61.0%
    • GPT 6 Luna: 78.3%Top

Scores from Artificial Analysis, updated Oct 26, 2026. OpenAI o4-mini high was tested at high reasoning effort; compare scores measured at the same effort.Artificial Analysis

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/o4-mini-high",
    "reasoning_effort": "medium",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="openai/o4-mini-high",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

OpenAI

Other OpenAI models, starting with the same family.